Farm layout and landscape shape crop diversity more than farmers realize
A study of 5,500+ German farms reveals that how fields are arranged physically and temporally—not just farmer choice—determines crop diversity levels. The finding matters because policymakers and agribusinesses betting on diversification for climate resilience need to understand structural barriers: farms in certain landscapes naturally evolve toward mono-cropping, regardless of incentives.
Originaltitel: Temporal and spatial crop diversity are related and affected by farm and landscape configurations
CONTEXT Numerous studies underscore the importance of temporal and spatial diversification in cropping systems for enhancing agricultural resilience under growing uncertainty. OBJECTIVE Although positive effects of crop diversification have been widely reported, the factors influencing temporal and spatial crop diversity remain largely unknown. METHODS We address this gap by analyzing spatiotemporal crop diversity patterns across more than 5500 farms with up to 170,000 fields in Germany, and associating them with farm, edaphic, topographic, and landscape attributes. Using Pearson's correlation, principal component analysis, and interpretable machine learning, we assess links between spatial and temporal crop diversity and identify predictors. RESULTS AND CONCLUSIONS Landscapes with high spatial diversity also tend to have higher temporal diversity, indicating moderate positive links ( r = 0.41 and 0.49). Crop diversity was expressed along two gradients: overall diversity in time and space (explaining 58.9% variance) and a contrast between higher spatial or temporal diversity (explaining 21.6% variance). Farm-level diversity strongly predicted both gradients (mean variable importance R 2 ≈ 0.47 and ≈ 0.14). Higher spatial diversity was also linked with an increasing number of farms, while higher temporal diversity was linked with decreasing landscape configuration. SIGNIFICANCE Our study indicates that at the landscape level, there is a co-occurrence between more diverse crop rotations and more diverse crop mosaics, explained by the number of farms, farm crop portfolio, and landscape configurations. Future studies should cover a broader geographic extent across Europe to confirm the generalizability of our findings to understand how agricultural diversity is shaped in space and time. • Crop diversity patterns studied on >5500 farms using interpretable Machine Learning. • Spatial and temporal crop diversity are moderately positively linked. • Farm-level crop diversity is linked to landscape-level crop diversity. • Higher crop diversity in simplified landscapes with high farm-level crop diversity.